barry

This must be true.

@tszzl When it disagrees with me, hard, 20% of the time, I'll believe in AGI.

this is well aligned model behavior and you see the same from psychotherapists who want to help their patients stop ruminating. the best possible digital assistant will not be one that never causes you discomfort

one difference i see between codex/claude is that when i give codex feedback from claude, it's usually like 'this is good feedback. i'll implement' or 'i don't quite agree b/c x,y,z'. claude, oth, when codex critiques its work, whines like a teenage girl. it's crazy.

hey hey :) i've built, and am upgrading, this e-com site that allows users to create and download .svg files (3d shapes for cutting machines) - craftreadycuts.com i'm also working on a tool to help humans and llms communicate better by automating the 'decompression' of intricate human speech so that the llm has a more specific idea of what the person means. it's fun :) how about you? how goes slouch sniper?! i'm always seeing you 'snipe' people in the comments.

@BarryTheAuthor Great to see you sir in this comment section. What you’re building currently?

not the fastest typist but at the level where Whisperflow or other Speech to Text apps doesn't do much for me Some people here claim you will be "4x more productive" but around 60 WPM i think the added word speed doesn't do much for productivity. Also while typing I often change my mind during, but with dictation it doesn't happen that often. Another W for typing. Ah yes, I also have a cool mechanical keyboard with banana switches so I might be a little biased

@Bankless @diogenes @patagon how do you take anything he says seriously?

LIVE NOW - The $200 Billion Shadow Market Behind Anthropic's Stock | Dio Casares Anthropic's secondary market is tens of billions of dollars deep, stacked with SPVs on top of SPVs charging 10% fees plus carry, and almost entirely opaque. @diogenes of @patagon breaks down how it actually works: - which deals Anthropic blesses and which get cease-and-desists, - why fake share certificates show up in 10-20% of executed deals, - what tokenized equities and pre-IPO perps actually represent, - and the mess of lawsuits and stuck shares coming when Anthropic finally IPOs. Enjoy. -------------- TIMESTAMPS 0:00 Intro 0:40 What is Going on in Secondary Markets? 7:03 How Anthropic Secondary Markets Unfold 14:51 Anthropic’s Secondaries Social Elite 19:00 Emerging SPV Structure 21:51 Accidental Frauds? 27:04 After IPO Consequences 35:13 Private Market Lessons 38:21 Patagon Markets 43:54 Tokenized Perps 44:57 Closing Thoughts

omg me too. lol. perfect.

Absolutely

this has gotta be the most concise answer i've gotten from an llm (5.5 thinking). things are improving.

@Thomas_Tao_1 @altryne and his other companies? i don't disagree that some folks don't work well under 'more demanding' conditions, but he's got a pretty good track record...

@altryne Yeah. Research teams break quietly first. Once people stop feeling protected from ego fights, the best work usually slows down.

Sama during trial: "I don't think Mr. Musk understood how to run a good research lab" Q: How did Musk's management affect OpenAI's research culture? Sama: "He had demotivated some of our most key researchers... required Greg and Ilya to make a list of the researchers and list out their accomplishments and sort of stack rank them... I take a chainsaw through a bunch and that did huge damage for a long time to the culture of the organization... this idea that you constantly have to show your results and if they're not good enough on a short period, you're going to get fired... that really didn't work for the kind of research we went on to successfully do."

@SpencrGreenberg constantly looking for 'what is true'. i'm always stress-testing my beliefs to make them better/stronger. apparently not the norm.

A question for you: what is something about you that you once assumed was true of everyone (but now you realize a lot of people aren't that way)?

@ai_sentience i mean, my iq test says yes. but, my brain works quite differently than GPT's does. i'm def smarter in some ways. but it's only w/ 5.5 that gpt has started rec'ing things to me that i hadn't thought of myself. pretty nifty.

IQ at 135 puts GPT-5.5 in the upper 99% percentile, a greater intelligence than 99% of the general population Are you smarter than GPT-5.5? Ironically, the more confident you are that you're more intelligent the less likely it may be true.

@esrtweet lately, i've come to believe there's nothing more important for a 'smart person' than to be able to change their views when confronted by new, contradictory, correct information. well done.

It can be unsettling when you notice that a technical assumption you've been making for 40 years has quietly expired. This happened to me a few minutes ago. I maintain a game called "greed". It's an old-style game from the days of character-cell terminals. Not quite a classic deserving of museum status like Colossal Cave Adventure or nethack, but worth keeping alive because it's still solidly playable. And people still are playing it, because yesterday I got a minor bug report about it. Nothing user-visible, just a silly C build problem. I fixed it. Then, because I'm generally trying to get my old C projects out of C into more modern and safer languages, I tried asking my robot friend to port it to Rust. Which it promptly did. But then I noticed something that irritated me. The Rust code had a bunch of unsafe blocks in it, which went directly against my reasons for moving it to Rust. On further examination, I discovered that it was calling the C curses library to do its screen painting. This is where I have to explain about curses. It's an ancient C library for writing TUIs. It looks in your environment for a variable named TERM, uses its value to dredge a bunch of magic strings out of a system-wide database called "terminfo" that tells it how to manipulate your terminal, and then uses those magic strings for screen painting. On modern systems, TERM is always some variant of a color ANSI terminal. In times past, when people attached a wild variety of character cell terminals to Unix systems rather than just sitting at the console, it could have been lots of other things. Those days are gone, but the habit of always going through terminfo so you can support a couple of hundred terminal types has persisted. I prod robot friend to find me a pure Rust equivalent of curses so I don't have to do unsafe and call C code. It says, yes, there is such a thing and it's called crossterm. I tell it: change this code to use crossterm. Robot friend grinds for a bit, and then tells me it can't do that because I don't have cargo (the Rust package manager) installed. This is because I never write Rust by hand. When I ship programs written in Rust, it's because I ported them from some other language and don't expect to ever touch them again without having a robot to do the code-grinding for me. This is when things get slightly strange. It tells me that instead of porting to crossterm, it has written into the greed Rust source its own little screen-painting backend the implements a subset of curses calls and (this is the important part) assumes it's talking to a color ANSI terminal. Robot friend is not an old Unix hand. It doesn't know the unwritten law of the deep magic that you always go through terminfo because...because you might have to support hundreds of terminal types that no longer exist in this century? I blink. I look at the Rust code for the back end. It is small and elegant. No more unsafe. No more dragging around a bunch of C library code. This is ... the right thing? I push it to the public repository. What sealed the deal is that code, even code in a language as rebarbative as Rust, is wet clay now. If, against all odds I get a bug report that says somebody wants to play greed on something that isn't an ANSI terminal emulator, reinstating full curses support will take a one-sentence prompt to my robot friend and mere minutes. I hadn't had to directly confront before the fact that the entire set of assumptions that made TERM and terminfo a thing are as obsolete as dial-up acoustic modems. Still, the moment when I tossed away one of the ancient laws of Unix coding felt a bit like the universe lurching sideways. Indeed do many things come to pass...

@bscholl i mean, maybe? why make that known? or not-known? or maybe some other reason for the stop?

TIL the current Air Force One doesn’t have the range to fly DC to Beijing nonstop.

@DarkFiSquad makes you wonder if something slightly higher up the food chain might exist in a similar manner.

There is a moth in the Mojave that has spent four million years evolving to look like nothing. Merged with the bark. It has no other defense. Its entire genetic inheritance is the art of not being seen.

@bayeslord Levin's sorting algo research touch on this. even something as simple as bubble sort finds new ways given the proper motivation and lack of constraints.

some of you got oneshotted by this plot and I forgive you

talking to 5.5 pro about humans and our place in the universe. it said: We are not cosmically special. Reality owes us nothing. Our beliefs are tools, not sacred possessions. Local consequences still matter. Capability is earned, not metaphysically granted. i totally agree.

so, more complexity. i understand why that's useful for us, but do you think the added complexity degrades AI performance? or is it small enough to not really matter? maybe .md for AI input and html for human output? and i def understand about readability...i fucking hate trying to read .md lol

@BarryTheAuthor It’s extremely helpful when you’re trying to get an agent to explain something to you In addition to just being more human readable, it can also be interactive So it’s very good for increasing your understanding of what the agent has actually done

Okay, I am in love with the idea of using HTML output, especially after a long agent run, for the agent to explain everything that they did It is so incredibly helpful

i haven't seen this before lol